Understanding Data on Recessions and Start-ups

In a recent post, Anita Campbell puzzled over the seeming contradiction between two recent publications – The Kauffman Foundation’s paper Exploring Firm Formation: Why Is the Number of New Firms Constant? (PDF) and the SBA’s paper Nonemployer Start-up Puzzle (PDF) — about whether recessions lead more start-ups. The SBA report showed that the employer business start-up rate is higher in  states with greater economic growth, while the Kauffman Foundation study showed that the number of new firms created annually is largely constant over time.  So, Anita was left wondering whether the economic environment influences start-up rates.

Some readers left comments on the site highlighting the fact that the studies contradict less than they appear to at first glance. As the readers pointed out, the studies’ results differ in large part because of differences in their data.  But none of the readers explained how these studies are less comparable than would first appear.  I don’t fault any readers for that – you need to look very carefully at the two studies to understand what they are doing before you can really understand their conclusions. So I am going to explain what I think are the differences between the two studies that account for their different results.

Rates Versus Levels

The SBA study measures rates of new business formation.  The authors divide the number of firms started by the size of the labor force. In contrast, the Kauffman Foundation study just measures the number of firms started.  This difference is important because the labor force has tended to grow over time.  And if the number of new businesses created annually stays constant and the labor force (and population) tends to grow, then the share of Americans who are starting businesses every year will decline over time.

In an earlier column I wrote on the New York Times small business page, I highlighted the fact that rates of entrepreneurship in the United States have been declining over time.  As the charts in that article show, if you take several of the measures of new firm creation shown in the Kauffman Foundation study and divide them by the U.S. population, you will observe a declining rate of new firm formation. That is, over time a declining share of the population is starting businesses.

It appears that the authors of the Kauffman Foundation study actually know this, but for some reason chose not to make that clear in their paper.  Buried in footnote 34 on page 17, they write Of course over this period of time, we’ve seen a declining rate of entrepreneurship, a phenomenon we will explore in a forthcoming paper.”  So basically, the Kauffman Foundation research shows that rate of entrepreneurship is declining over time because the number of new firms is constant and the population and labor force are growing.

Employer Versus Non-Employer Firms

The other big difference between the two studies is what they measure. The SBA study looks at both employer and nonemployer firms, while the Kauffman Foundation study focuses on employer firms.  (Non-employers are companies with at least $1,000 in revenues but have no employees other than the owner.) However, non-employer firms tend to be smaller than employer firms, but they make up three quarters of all firms in the economy and close to 80 percent of all start-ups.

In another New York Times column, I highlighted the fact that trends in the rate of creation of employer and non-employer firms are very different.  The rate of non-employer firm formation has been increasing in recent years, while the rate of employer firm creation has been decreasing.  These different patterns suggest that external economic conditions might affect the formation of the two types of firms very differently.

Moreover, the two types of firms appear to be different kinds of companies, rather than different stages in the lives of businesses.  Few non-employer firms “grow up” to become employer businesses. In a paper entitled, “Measuring the Dynamics of Young and Small Businesses: Integrating the Employer and Nonemployer Universes,” Steven Davis and his colleagues found that only three percent of non-employer businesses transition to employer businesses when observed over a three year period, and these businesses only account for 28 percent of young employer firms.  Thus, Davis and his colleagues conclude, “It is tempting to think of the nonemployer business universe as a vast nursery for employer businesses [from which] many nonemployers evolve into employers and a few eventually grow into giant corporations that generate thousands of jobs. However, as our results confirm, most nonemployer business are quite small and never become employers.”

Other research also shows that employer and non-employer firms are very different. Analysis by Rick Boden and Al Nucci, for instance, show that 85 to 90 percent of all new non-employer firms are sole proprietorships, a much higher percentage than the share of employer firms.  In fact, Davis and crew write in their paper, “Indeed, it is misleading to think of all records in the nonemployer universe as ‘businesses’ in the usual sense. Many nonemployer records reflect side jobs, hobby businesses or occasional consulting engagements that generate extra income for households that depend primarily on wages.”

This leads to another important difference between employer and non-employer firms, which is directly relevant to the comparison of the SBA and Kauffman Foundation studies.  People might be more likely to start employer firms to pursue business opportunities, while they might be more likely to found non-employer firms as a reaction to bad economic alternatives.  Examining the differences in the rate of formation of employer and non-employer firms across states, the authors of the SBA study found that employer start-up rates are positively correlated with real GDP growth, while nonemployer start-up rates are not related to economic growth.

Other Differences

There are also some other differences between the studies that might affect their findings.  The Kauffman study doesn’t conduct any statistical analysis to net out the effect of forces other than economic conditions on start-up activity, while the SBA study controls for these other effects.  The SBA study looks at differences between states at a point in time, whereas the Kauffman study looks at differences in the country over time.

In short, the two studies don’t tell a different story about what happens to entrepreneurship in response to different economic conditions as much as they explain the difference between employer and non-employer firms, and the difference between rates and levels of entrepreneurship.

More in: 16 Comments ▼

Scott Shane Scott Shane is A. Malachi Mixon III, Professor of Entrepreneurial Studies at Case Western Reserve University. He is the author of nine books, including Fool's Gold: The Truth Behind Angel Investing in America ; Illusions of Entrepreneurship: and The Costly Myths that Entrepreneurs, Investors, and Policy Makers Live By.

16 Reactions
  1. Scott mentioned this important fact (among others):

    “Examining the differences in the rate of formation of employer and non-employer firms across states, the authors of the SBA study found that employer start-up rates are positively correlated with real GDP growth, while nonemployer start-up rates are not related to economic growth.”

    While basically true, it is not 100% true. It misses the mark of what I consider a significant source of economic growth: The Virtual companies that are popular in the wake of the “dot com meltdown.” In 1998, I formed an LLC that generated a small return personally, but fueled multimillion dollar revenue for my clients. Some of us are more interested in making money than simply hiring and managing and trying to compensate employees.

    By using subcontractors for 100% of the labor, I paid only for results. Something that other business owners are learning. Both traditional and virtual.


  2. What a wonderful wealth of knowledge! Thanks for sharing.

  3. Hi Scott, Thanks so much for taking the time to explain the difference between those two studies. We really appreciate your expert insights and this explanation.

    Although I can’t let pass without comment the quote of the authors of the 3rd study you mentioned, without challenging them on this statement they made:

    “Many nonemployer records reflect side jobs, hobby businesses or occasional consulting engagements that generate extra income for households that depend primarily on wages.”

    Let me offer 3 points to their comment:

    (1) They threw in that conclusion gratuitously at the very end of their paper. As a former lawyer, I can tell you if they’d been in a court and made that statement the opposing attorney would have been jumping up and down yelling “Objection! Hearsay! No foundation!” and the judge would have sustained the objection and thrown their statement out of court in a New York minute.

    (2) While I would agree that employers and nonemployers often are two different universes, I would submit that a study needs to monitor businesses beyond 3 years. As someone who’s started a bootstrapped business, I can tell you 3 years is not very long. I strongly suspect the picture would look much different over a longer horizon. That’s what bootstrapping is all about — it’s slooooow growth.

    (3) I would reiterate Ski’s point about virtual businesses. Implicit in that paper is the suggestion that economic value is pretty much limited to jobs. In today’s world of “virtually organized” businesses they completely miss the economic picture of what’s going on by focusing on jobs. If somebody wanted to do a study that was REALLY helpful to the economy, they’d figure out a way to measure the economic impact of those nonemployer businesses through their spending on business products and services.

    But I suggest that the really valuable research is being done by corporations that sell to these nonemployer businesses — they understand the value in the economic stream.

    Sorry for getting on a soapbox — but I couldn’t resist responding to that statement contained in that third paper, Scott. 🙂

    – Anita

  4. Thank you Scott!

    One other thing that’s not mentioned, and it’s this;

    The personality traits of the owners of non-employer start-ups, vs. start-ups that actually have employees early are most likely quite different.

    It may be that many of the one-person businesses out there started out as one-person businesses because of their personality traits, and even their professional skills. The folks behind them may not be business “builders.” A one-person business is usually easier to manage, and very flexible. One can do quite well in one, too.

    A business that has employees vs. one that doesn’t are two different beasts. They require different mindsets, and very different people.

    The Franchise King

  5. Jinx! 😉

    Martin Lindeskog says:
    January 18, 2010 at 3:15 pm

    Anita: Have you asked Scott Shane about his take on the data?


    • Yes, Martin, he responded to your request! How’s that for being responsive, huh? We’re lucky to have Scott’s views shared here. It’s makes for really interesting conversation.

  6. I would add that many non-employer start-ups also avoid adding employees due to the regulatory & tax burden. It’s just a lot easier that way.

  7. More and more people are just working for commissions so that is why the trend in consultants has risen.

  8. Anita: Yes, that’s great. Thank you very much Scott & company. 🙂

  9. Great post and thread!

  10. I followed this topic and can appreciate the two studies may appear as comparing apples to oranges thus a contradiction between them. The two studies might have not been compared in the first place.

    The question remains do recessions lead to start ups? Surely there are stats from past recessions and studies that measure recessionary periods to begin the discussion. To answer the question, recessionary periods need to be compared and conclusions need not be based formally on job growth over time as related to population, labor force, or GDP in general.

    To say non employer businesses are not defined in the strictest sense as “business” is subjective at best. All that means is non employer business do not render data nor has it ever been as popular as it is in the current recession. Tax officials have hit a ceiling because payroll taxes (as well as insurance)are unaffordable. Employer businesses reduce payroll and shift to a model that can increase independent contractors legitimately; new start ups are based upon the hiring of independent contractors.
    I agree with Ski on this post in terms of independent contractors and virtual models.

    I am not surprised non employer firms are rising, particularly in the current recession. Necessity is the Mother of Invention and people need more than they want presently.

    I challenge the comment by the author that non employer business does not relate to economic growth. The fact no data exists to study non employer business in a recession (or over time) does not translate into its having no direct impact on economic growth.

    It is to the benefit of the owner of a start up business in a recession not to be defined by authorities, not to be traced by researchers, and not to be tied directly to the national economy; inherent in the lack of definition lies the definition of entreprenuership. If the focus remains upon employer business, its data, traditional formulas inherent to economic growth, and upon global trends not specific to recessions historically, there will never be a satisfactory answer to the question to do recessions lead to start ups? The question may be asked from the top down but can only be answered at the ground level.

    I founded a non employer business two years ago. My business is virtual and provides accounting service to small business start ups and small businesses in transition. I see first hand the growth of non employer businesses and the decline of employer businesses in a recession. As long as there are two economies in America…Wall Street and Main Street…expect Main Street to survive without Wall Street. But I dare say Wall Street can’t really survive without Main Street because the former depends too much upon employer business revenue.

  11. Whenever I see mention of data from Kauffman I get nervous. Kauffman is a premier source of information on entrepreneurship and I applaud their work BUT when I was researching my book I encountered discrepancies between their statistics and almost everyone else I consulted, including the SBA. BIG discrepancies. Example: if Kauffman data on start-ups is accurate then a comparison of their stats on start-ups and the actual number of ongoing businesses in the US at any given time would indicate a start-up failure rate of over 95%. This is far greater than anyone else asserts or assumes.(The consensus figure, which I use in my book, is 80%. See my Chapter One,footnote 12.) Two guys sitting in a bar nursing a beer and grousing about their job and how they would like to be out on their own with no boss does not constitute a start-up. The same is true for someone trying to sell some junk from their attic on eBay. Re: Kauffman’s recent study, common sense (as well as my own experience in and observations of small business and franchising) tells me that the number of self-employed “start-ups” increases (out of necessity) but the number of start-ups that employ workers declines during most recessions. As the economy kicks in, these trends reverse themselves.